The present invention relates to the art of heat sinks and cold plates. It finds particular application in conjunction with electronic circuitry used in industrial variable-speed electric motor drives and will be described with particular reference thereto. However, it will be appreciated that the present invention will also find application in conjunction with other electronic devices including non-industrial electronic devices and in any other application which requires a heat transfer or exchange. As an example, the present invention is well suited for the automotive industry where the cooling of electronic and other heat generating components under the hood of an automobile is carried out using readily available refrigerant or radiated closed-loop engine water coolant in conjunction with the invention. Also, the present invention is applicable to provide a specialized electric motor housing adapted to circulate a cooling fluid through the motor casing.
It is well known that variable speed drives of the type used to control industrial electric motors include numerous electronic components. Among the various electronic components used in typical variable-speed drives, all generate heat to a varying degree during operation. Typically, high-power switching devices such as IGBTs, diodes, SCRs, capacitors and the like are responsible for generating most of the heat in a variable-speed drive. It is for this reason, therefore, that most variable-speed drives include a heat sink(s) upon which the power switching devices are mounted. The heat sink(s) conducts potentially damaging heat from those components.
Selecting the size and design of a heat sink for a particular variablespeed drive is somewhat of a challenge. First, a designer must be aware of the overall characteristics of the motor and drive pair. Next, the designer must understand the industrial application into which the motor and drive pair will be used, including the continuous and peak demands that will likely be placed on the motor and drive by the load. Third, the designer must accommodate, in the design, certain unexpected conditions that would deleteriously affect the heat transfer capability of the heat sink such as unexpectedly high ambient temperatures, physical damage to the heat sink such as mechanical damage, or a build up of a debris layer, as examples. Lastly, the heat sink(s) must be physically dimensioned so as to fit into the space allotted per customer requirements cabinet or enclosure size, or the like.
In the past, air-cooled heat conducting plates were used to transfer thermal energy from electronic parts to the ambient air. These were passive heat-transfer devices and were generally formed of a light-weight aluminum extrusion including a set of fins. As a general rule, heat transfer effectiveness is based on the temperature differential between the power devices and the ambient air temperature. Of course, in order to provide adequate heat conduction, heat sinks of this type oftentimes are necessarily large and, therefore, bulky and expensive. If high ambient conditions exist, the heat sink becomes ineffective or useless as heat removal cannot be accomplished regardless of the size of the heat sink. If the variable speed drive was in an enclosed space the heat removed from the drive would need to be exhausted or conditioned for recirculation.
By forcing air over fins defined on the heat-conducting plate or aluminum extrusion, improved cooling efficiencies were realized. Large blower motors are often used for this purpose. However, as the fins defined in the aluminum extrusions become dirty or corroded during use, the heat sinks become less effective or useless altogether. Blower motors cannot be used in environments where air cleanliness would clog filtration. Therefore, air conditioning equipment is often added to internally circulate and cool the air that is passed over the heat sink fins.
Independently cooled cold plates have also been used for some applications but with limited success. Because of their cost, both in the construction of the cold plate itself and in the additional peripheral support apparatus required such as fluid pumps, conduits and the like, heat sinks of this type are typically used only in applications which require a critical degree of control over the temperature of the electronic drive components or where ambient air temperature is excessive.
Once such independently cooled cold plate system is described in U.S. Pat. No. 5,523,640 to Sparer, et al. The Sparer, et al. '640 patent teaches a cooling system which circulates a liquid coolant through a plurality of specialized motor housings and through a cold plate providing on one side a mounting surface for electronic drive components. The multiple motors in the Sparer, et al. '640 system are provided with a housing having an integral heat exchanger. The various electronic components forming the motor drives are mounted to the surface of a chill plate. A housing structure which typically resembles a box and which encloses the electronic components is provided to protect the components. The heat exchangers in the motors are formed by casting stainless steel tubing into a cylindrical aluminum stator housing. Similarly, the chill plate is made of stainless steel tubing cast into an aluminum plate-type heat exchanger.
While the Sparer, et al. '649 system is a marked improvement over forced air type heat exchangers, such independently cooled cold-plate heat exchangers are difficult and costly to produce repeatably. This is due mainly to the inescapable requirement to use stainless steel tubing rather than copper tubing in the aluminum casting process. Stainless steel is better able to withstand the molten aluminum. However, stainless steel is difficult to bend and machine. Stainless steel is also expensive. Bending may require annealing of material which weakens the tubing construction. The weakened annealed areas are prone to aluminum "blow through." Stainless Steel will also develop surface corrosion when exposed to typical coolant chemicals or chemical compounds which attack the steel's passive protective surface. This corrosion reduces the steel's ability to dissipate heat.
However, as those skilled in the art would appreciate, stainless steel tubing is used in chill plate construction because it is one of the few materials able to withstand exposure to molten aluminum and other similar molten metals in a mold without normally developing "blow holes" during the casting process. As the holes develop in metals other than stainless steel, they permit molten pour material to enter the tubing where it solidifies forming a blockage. This renders the heat sink useless. Blow holes do not normally form in standard temper stainless steel tubing because it does not alloy with molten aluminum. Also, the liquidus of stainless steel is much greater than the liquidus of aluminum.
Although it would be desirable to use tubing formed of a material other than stainless steel, attempts to repeatably manufacture aluminum or copper heat sinks using aluminum or copper tubing have heretofore failed. Prior to the present invention, casting with copper tube in an aluminum plate had been attempted but could not be repeated for production purposes due to the above described alloying difficulties. In the past those attempting to cast copper tube in aluminum have had yield rates of one out of ten or 10%.
In addition to alloying and associated blow through problems, copper tubing can also, under certain circumstances, produce dangerous conditions during a molding process. For example, at a specific temperature copper is known to re-crystallize (i.e. change its crystalline structure). Upon recrystallization copper gives off a gas during an "outgassing" process. In the case of a sealed mold form, trapped heated gas expands and increases form pressure. If form pressure exceeds a peak form pressure, the form and molten metal therein has been known to explode causing a bomblike effect sending molten metal and form shrapnel in all directions. Obviously a pressure filled form presents a dangerous situation. Therefore, for this reason also, copper tubing has been avoided in sealed molding processes.
Moreover, while copper tubing can be manipulated more easily than stainless steel, some tubing configurations such as tight tubing radii cannot be provided via tube manipulation. Tight radii has to be provided by either constructing special tube forming equipment (i.e. a special extruder) or by securing prefabricated conduit components (i.e. elbows, "T" sections, straight sections, etc.) together. Special equipment is prohibitively expensive.
While prefabricated conduit components can be brazed together to form a suitable conduit construct, brazed joints have at least one primary disadvantage. Copper-zinc and copper-sliver are typical brazing compounds. After a brazing compound hardens and secures adjacent components together, like copper tubing, brazing compound alloys with molten aluminum and out gasses when a specific temperature is exceeded. The brazing compound, unfortunately, alloys with aluminum at a lower temperature than the copper tubing. In addition, the brazing compound out gasses at a lower temperature than the copper tubing. Thus, when a conduit construct is formed via brazing, the yield rate is further reduced by increased blow through and the possibility of a form explosion is increased due to additional compound outgassing.
With respect to stainless steel, stainless steel tubing is not only difficult to work, as discussed above, but its thermal resistance is higher than aluminum or copper. The stainless steel tubing string in an aluminum or copper casting acts as a thermal insulator as compared to the heat transfer characteristics of the aluminum or copper forming the heat sink body. It would be desirable to use a tubing material having thermal transfer characteristics that match the aluminum or copper forming the body of the heat sink. It would be preferred to use aluminum or copper tubing in an aluminum or copper heat sink. Attempts to repeatedly manufacture such heat sinks have heretofore failed.
Much development has occurred in design of specially formed and/or machined tubing to increase its heat transfer capability. Tubing has been manufactured in various forms to increase heat removal by increasing flow turbidity and cross-sectional area. As an example, "gun barrel" type tubes and knurled tubes have been used. In the past, attempts to cast these type tubes in aluminum or copper heat sink bodies have generally failed as well. The present invention is the only known method for repeatably casting such tubing forms with high yield rates.
An additional limitation of prior independently cooled cold plates such as the system taught in the Sparer, et al. '640 patent discussed above is the heat absorption capacity of the coolant fluid. Typically, the fluid used is water or a glycol water solution. In heavy industrial applications, the cooling water is typically circulated between the heat sink and an auxiliary heat exchanger such as a liquid/liquid-type unit. The auxiliary heat exchanger is usually connected to a source of tap water or to a central water chiller. This additional heat exchange equipment is costly, complicated, requires periodic maintenance, and is prone to failure.
With any type of cold plate design tube spacing is extremely important for a number of reasons. First, as a general rule heat dissipation is a function of how close tubes are to the surfaces of a heat sink and more specifically how close the tubes are to the heat absorbing surfaces.
Second, where similar devices which generate similar amounts of heat are mounted to a cold plate surface, to ensure essentially even heat dissipation the tubes must generally be equispaced within the plate.
Third, where different devices which generate different amounts of heat are mounted to a cold plate surface, to dissipate different amounts of heat at different locations on the plate, the tubes must be differently spaced in different portions of the plate. For example, where power devices give off more heat than capacitors, relatively more tube length should be formed within a plate portion adjacent the power devices than in the plate portion adjacent the capacitors.
Fourth, where electronic devices are to be attached to a heat sink via bolts received in holes, it is extremely important that, during hole placement, tube location be precisely known. If tube placement is unknown, a bolt hole may be provided which enters a tube and thereby renders the sink assembly inoperable.
Fifth, other machining may trim the thickness of a wall after molten sink material cures. In this case, if tube position is not precisely known, the machining may form a relatively thin wall adjacent a tube or, in fact, may enter a tube.
Unfortunately, while precise spacing is important, often, during heat sink formation, spacing cannot be precisely maintained. While stainless steel can be formed into relatively rigid serpentine tubes prior to providing molten sink material therearound, when molten material is added to a sink mold, the material tends to force the tubes vertically and, in some cases, horizontally, within the mold cavity. This movement changes the position of the tubes with respect to each other and with respect to the finished sink surfaces.
While tube movement is problematic even in the case of stainless steel tubes which are relatively rigid, it is likely that such movement would be exacerbated in the case of tubes formed of a softer and less rigid material such as copper or aluminum.
One attempt to ensure tube spacing during mold formation is described in U.S. Pat. No. 5,484,015 which is entitled "Cold Plate and Method of Making Same" which issued on Jan. 16, 1996. That patent teaches the use of tie bars which are used to maintain the vertical spacing of tubes within a sink during a molding process. To this end, adjacent tubes are stacked together (i.e. touch each other) and tie bar is secured therearound. In this configuration there is no space between the stacked tubes. Top and bottom portions of the tie bar serve as vertical spacers for the tubes when the tube assembly is placed within a mold cavity.
Unfortunately, while the tie bars described in the '015 reference can in fact maintain vertical spacing between mold surfaces and adjacent tubes, the bars cannot maintain either horizontal tube positions or positions of tubes with respect to other tubes. For example, the reference does not teach how to, within a sink wall, maintain a one inch space between first and second tubes or a one inch space between each tube and all adjacent wall surfaces.
Although most electronic devices generate heat from virtually all device surfaces, devices are typically designed to generate most heat from a single device surface. For the purposes of this explanation the surface of each device which generates the most heat will be referred to hereinafter as a dissipating surface.
Prior art cold plates have included electronic devices secured along their dissipating surfaces to only a single cold plate mounting surface. Mounting to a single mounting surface has the advantage that all device input and output connections are exposed in proximity to each other. In addition, by limiting devices to single mounting surface, the opposite plate surface can be used to mount the plate and device assembly to some type of support mechanism (e.g. a wall or the like). With devices arranged on a single mounting surface, the mounting surface area required to mount the devices is equal to the combined surface area of all the device dissipating surfaces plus some clearance area between adjacent devices.
Unfortunately, while mounting to a single mounting surface has certain advantages, such a configuration can result in an overall assembly which requires a relatively large volume. This is particularly true where electronic devices have disparate shapes and sizes.
For example, as well known in the motor controls art, to configure a motor drive to store DC voltage and then convert the DC voltage to variable AC voltage, two general device types are required including (1) storage capacitors for storing the DC voltage and (2) some type of power switching devices (e.g. IGBT, BJT, GTO, etc.) for converting the DC voltage to variable AC voltage. While a motor control assembly typically includes six switching devices and a large number (e.g. 36) of capacitors, in order to explain how devices which are mounted to a single plate surface result in assemblies which require large volumes, it will be assumed that a configuration includes only one capacitor and one switching device.
Capacitor and switching device shapes are typically very different. Referring to FIG. 11, a plate assembly A1 including a single capacitor C and a single switching device S which are mounted to a single mounting surface 207 of a plate P is illustrated. Capacitor C shape is generally elongated and cylindrical, having a length Lc and a width Wc, and each capacitor has a single relatively small dissipating end surface 205. To facilitate maximum heat dissipation the capacitor's heat dissipating surface 205 is mounted to a cold plate mounting surface 207.
Unlike a capacitor, most switching devices S have a generally flat configuration having a thickness Ts. The largest surface of a switching device S is usually the dissipating surface 209 which has a length Ls (not illustrated) and a width Ws. Typically each of length Ls and width Ws is much greater than thickness Ts. The switching device dissipating surfaces 209 are mounted to the mounting surface to facilitate maximum heat dissipation.
Referring still to FIG. 11, plate P has a thickness Tp. A simple housing H is provided to protect capacitor C and device S. The total volume V1 required for assembly A1 is: EQU V1=(Lc+Tp) (Wc+Ws)(Da) (1)
where Da is assembly depth into FIG. 11 (e.g. where assembly A1 is one device deep, Da may be either device S length Ls or capacitor C width Wc). Clearly, a large volume Vw between device S and adjacent portions of housing H is unpopulated (i.e. has no device located therein) and therefore, that volume V.sub.w is wasted.
One way to reduce the volume V1 required for assembly A1 would be to provide a relatively complex housing structure which conforms to the shapes defined by the mounted capacitor C and device S. Unfortunately, such a housing would be relatively expensive to configure as various contorted shapes would have to be accommodated. In addition, where another capacitor is added to FIG. 11 adjacent device S and on a side opposite capacitor C (i.e. to the right of device S), the complexity of a volume saving housing would be exacerbated and not very effective in any event.
While wasted volume Vw in simple configuration A does not seem appreciable, the wasted volume increases as additional devices are added to the assembly. For example, as additional devices are added to assembly A1, volume V1 increases by at least the area of the dissipating surface of the added device times capacitor length Lc. The additional wasted volume is appreciable.
Moreover, where an assembly A1 includes a large number of devices mounted to a single surface, the shape of the housed assembly often has a relatively large footprint (i.e. width and length sufficient to accommodate all dissipating surface areas plus clearance between devices). While an assembly having such a large footprint may be acceptable in some applications, such an assembly will often be prohibitively large for use in other applications. For example, in an electric vehicle where space is limited and often has contoured boundaries, assemblies having large footprints often cannot be accommodated.
Yet another problem with cold plate designs is that the relatively long serpentine tube required to guide coolant through the plate impedes fluid flow therethrough to some extent. For this reason, often relatively large pumps are required to ensure sufficient pressure to pump the coolant through the tube.
One solution which would reduce tube impedance and thereby facilitate use of a relatively small pump would be to provide a manifold which splits coolant flow into several different paths through the plate thereby reducing tube pressure. This solution has been avoided for a number of reasons. First, to link an inlet tube to a manifold and, perhaps to form the manifold itself would require brazing together two or more different tube sections. With respect to stainless steel, brazing is an extremely difficult process which requires a large amount of skill.
While brazing aluminum and cooper components together is easier than brazing stainless steel components together, as indicated above, there has never been an effective way to form a sink about such tubing successfully. In addition, it should be noted that, even if a sink could be formed using copper or aluminum tubing, generally, the industry has viewed such brazed joints as unsafe.
Therefore, it would be advantageous to have an apparatus which could be used during a sink molding process to maintain both the horizontal and vertical positions of sink tubing, to have a sink which requires a relatively small pump and provide a sink and associated electronic components which together require a relatively small volume.